You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We want to only do the minimum amount of work necessary to properly group an event, and to display its primary / main thread stack trace.
We assume that the number of users that actually look at a background thread stack trace is very low. It should be possible to skip a ton of work, and do that work on-demand when a user switches to a background thread in the main event UI.
This work includes:
Symbolication and source context application.
Enhancement rules that categorize the stack frames and influence how it is displayed in the UI.
Apart from avoiding a ton of processing resources on things that users are extremely unlikely to ever look at, we would also save on storage costs as those parts of the event payload would only be backfilled on demand.
Doing this however is a high-effort initiative as it touches on a lot of places in the whole pipeline. Its success is also dependent on:
Availability of minidumps, as they are needed to retroactively stack walk background threads if that is not done eagerly.
Availability of debug symbols and source maps. As we are currently thinking about / planning for newer uploads to artifact bundles overwriting previously uploaded files, this means that older events might not resolve the same with newer debug files.
The text was updated successfully, but these errors were encountered:
Related to #51517 and #51518. And also getsentry/rfcs#106 eventually.
We want to only do the minimum amount of work necessary to properly group an event, and to display its primary / main thread stack trace.
We assume that the number of users that actually look at a background thread stack trace is very low. It should be possible to skip a ton of work, and do that work on-demand when a user switches to a background thread in the main event UI.
This work includes:
Apart from avoiding a ton of processing resources on things that users are extremely unlikely to ever look at, we would also save on storage costs as those parts of the event payload would only be backfilled on demand.
Doing this however is a high-effort initiative as it touches on a lot of places in the whole pipeline. Its success is also dependent on:
The text was updated successfully, but these errors were encountered: